MarcoDewey

9 ms·

121.
Launch HN: Jazzberry (YC X25) – AI agent for finding bugs
46 points by MarcoDewey 1y ago|34 comments
122.
AI is a test of our intelligence(psychologytoday.com)
8 points by MarcoDewey 1y ago|0 comments
123.
by MarcoDewey 1y ago
What if - some of us - just like to use excessive - dashes when writing -- ?
124.
Google tests replacing 'I'm Feeling Lucky' with 'AI Mode'(techcrunch.com)
4 points by MarcoDewey 1y ago|0 comments
125.
PhD students bring AI Coaching to Japan(miragenews.com)
1 points by MarcoDewey 1y ago|0 comments
126.
Pope addressed AI as a matter of concern(apnews.com)
5 points by MarcoDewey 1y ago|0 comments
127.
Chegg to lay off 22% of workforce(gazette.com)
5 points by MarcoDewey 1y ago|1 comments
128.
Using AI to predict biological age(washingtonpost.com)
2 points by MarcoDewey 1y ago|0 comments
129.
AI in Baseball Training(techxplore.com)
1 points by MarcoDewey 1y ago|0 comments
130.
by MarcoDewey 1y ago
What are some of your best prompting practices to get AI to write reliable code? I have been experimenting with forcing the AI to write the test cases first and then build software that passes all those tests.
131.
by MarcoDewey 1y ago
I love that you all are doing real old school machine learning and not just LLM transformer based work!
132.
Bug Finding Bot vs. Gruyere(jazzberry.ai)
3 points by MarcoDewey 1y ago|1 comments
133.
by MarcoDewey 1y ago
I believe that the unprecedented scale of LLM-generated code will demand a novel approach to software review and testing. Human review may not be able to keep up (or will it become the bottleneck?)
134.
by MarcoDewey 1y ago
I think this is a seriously excellent point. The bet that I am making is that the system reduces its error rate by splitting a broad task into two more focused tasks. However, it is possible that generating meaningful test cases is a harder
135.
by MarcoDewey 1y ago
improving code generation would be awesome :)
136.
by MarcoDewey 1y ago
I believe that I have unintentionally misled you. When I say "unbiased oracle" I am talking specifically about the test oracle being unbiased by how the software was implemented. ie. Black Box testing. I don't think I made th
137.
by MarcoDewey 1y ago
Yes, this test-driven approach will likely increase generation time upfront. However, the payoff is more reliable code being generated. This will lead to less debugging and fewer reprompts overall, which saves time in the long run. Also agr
138.
by MarcoDewey 1y ago
You are correct that the notion of LLMs being completely unbiased or neutral does not make sense due to how they are trained. Perhaps my title is even misleading if taken at face value. When I talk about "unbiased oracles" I am sp
139.
LLMs as Unbiased Oracles(jazzberry.ai)
34 points by MarcoDewey 1y ago|27 comments
140.
Testing in the Age of LLM-Generated Code(jazzberry.ai)
2 points by MarcoDewey 1y ago|0 comments
141.
Testing Code Means Running It(jazzberry.ai)
3 points by MarcoDewey 1y ago|0 comments